Available online at www.sciencedirect.

com

Procedia - Social and Behavioral Sciences 00 (2014) 000–000

www.elsevier.com/locate/procedia

XI Congreso de Ingeniería del Transporte (CIT 2014)

Travel Estimation Model Generated by Public and Private Schools,

According to Different Transport Modes. Case Study Mérida-
Venezuela
Angela Quintero Petit; Mary Isabel Díaz; Emilio Moreno González *
Universidad de Los Andes, Nucleo La Hechicera, Mérida-5101, República Bolivariana de Venezuela

Abstract

In many studies of transportation is required to estimate the trips intensity that generates certain land use, in order to measure
their impact on adjacent streets. Typically, when the data cannot be obtained directly, or by planning approach, parameters
suggested by the Trip Generation Handbook (TGH) of the Institute of Transportation Engineers (ITE) are used. However, the use
of these parameters in places outside its application scope might not be appropriate. Therefore, this work proposes a research
aimed to obtain trip generator models for different land uses, through random sampling, in schools of the Libertador
municipality, in Mérida-Venezuela. Evaluated transportation modes are: private car, school transport, public transport,
motorcycle and travel on foot. Data collection methodology presented a redundancy problem, due to quantify trips made by
different people in the same vehicle; this requires a change in future surveys. However, the incorporation of the occupation
indices and statistical analysis help to quantify the trips by mode making inference to the population. The influence of the socio-
economic level in the mode choice is evidenced, resulting useful trip generation rates that can be compared and incorporated into
the Red Iberoamericana data base as a contribution.
© 2014 The Authors. Published by Elsevier Ltd.
Selection and peer-review under responsibility of CIT 2014.

Keyboard: Trip generation rates; modes of transportation

1. Introduction

Trip generation rates allow the prediction of land-use behavior, specifically implemented in a certain place
depending what was observed in the existing land uses of similar characteristics, i.e. allows foreseeing the impact

* Corresponding author. Tel.: +58-274-2402877; fax: +58-274-2402877.

E-mail address: angelaqp@ula.ve

1877-0428 © 2014 The Authors. Published by Elsevier Ltd.

Selection and peer-review under responsibility of CIT 2014.
2 Quintero, Díaz &Moreno/ Procedia - Social and Behavioral Sciences 00 (2014) 000–000

that it will have on adjacent street prior to implantation, which is very useful to implement measures that mitigate
potential impacts. Trip generator data allows determining specific accessibility requirements, estimating the future
traffic intensity, and supporting proposals for transport systems improvements, evaluating the impact of zoning
changes or land uses.

Among the major limitations found in Venezuela to make these demand forecasts by land use, is the lack of data
that allows an adequate estimate. As a result, is much consulted the rates source by Institute of Transportation
Engineers (ITE), estimated with data not necessarily applicable to our cities. On the other hand, the local adaptation
sometimes requires data acquisition on existing infrastructure similar to the study project, representing greater cost
in the development of the project.

The Trip Generation Handbook (2001) recommends studies when it is required to establish local trip generation
rates, thus the Andes University roads department, in Venezuela, has initiated a project to obtain trip generation rates
to different land uses, which can be used. However, a greater effort is required to consolidate a wide database
applicable to local environment and that spans a large number of land uses still under study. The first results of this
research are models that can be used to estimate traffic intensity by transport mode, applicable to the population of
schools, as a first step, in database obtaining.

Thus, it is possible to determine trip generation rates relative to twenty-one schools; of which eleven are private
and ten are concerted –middle-payment–, located in the Libertador municipality of Mérida in Venezuela. Surveys are
applies to students and employees of each school, which allows to obtain different trips correlations according to
different variables, following the format of the Institute of Transportation Engineers, where average, range, and other
associated statistical parameters, they are shown. This data allows determining the trip intensity generated by each
transport mode according to its percentage distribution as a work objective. The results could be used to predict
adequate number of trips by class of school, incorporating the occupation indices as a necessary adjustment, and also
to correct the data redundancy, efforts to improve the data collection methodology in future work should be.

2. Background

Herz, Galarraga & Pastor (2009) made the trip characterization in universities of Córdoba city, Argentina. They
studied thirteen generators-poles and formulated modal distribution models for motorized individual, mass and non-
motorized cases. They used to students and employed as explanatory variable, and considered the generator-pole
location in the urban area. They made a first data collection on a Córdoba Catholic University campus with unique
access, where vehicle and people are registered, discriminating, car, bus, and others mode under entry and exit
conditions. They also used data collection through surveys and censuses, and they observed that the modal choice
varied significantly according to the University, considering the population density close to University and the socio-
economic characteristics of the students, to be able to explain the modal choice. People trip generation models were
statistically adjusted per day, depending of the schedules. In particular, four models were calibrated: total trips
considering all modalities; trips in individual motorized means: car, taxi and motorbike; trips by public transport;
and trips by non-motorized means, considering the bike and walk.

Souza & Prudencio (2010) on the other hand studied middle school institutions in Brasilia, concluding that,
although the issue has not been sufficiently explored in the literature, trips generated by study activities have an
important impact on individual trip, which represents the main reason for displacement. The data collection surveys
were applied to students, teaching and administrative staff, where the results show significant differences between
the trip generation of public and private schools. Observed characteristics can contribute to a better planning of the
vehicles movement and pedestrians around school, favoring safety, and providing trip generation models most
appropriate for this generator-pole.

A procedure for the data acquisition in schools is developed (Prudencio et al., 2006) where it emphasizes the
importance of the trip generation motorized and non-motorized study focus to schools for purpose of transportation
 Quinterot, Díaz & Moreno/ Procedia - Social and Behavioral Sciences 00 (2014) 000–000 3

planning and management of traffic in urban areas, since these institutions represent important trip generators-poles.
Also the Ibero-American network of trip generators-poles including schools, so the results may add to this great
database. Thus, a standard procedure for data collecting that included instruments for data registration is proposed;
where a better way of applying these instruments is defined, and presents a basic model that facilitates registration
and results consolidation obtained in field. This process consists: data collection in school through surveys -
appropriate to each school and population class- and processing of data collected in a common database with a
predefined format. The proposed surveys for schools and universities vary in its application to students, teachers and
employees. Through this proposal, it is possible to identify the characteristics of teachers, students and employees
trip by transport mode and for all educational levels, also to determine parking area, main routes for access to the
institution, and create alternative of different transport means. Two base questionnaires have a form that can be filled
by the investigator with the information given by each school head, raising general data such as; study type, land and
construction area, room number, parking zone, class schedule, and number of people -teachers and students-.
Questionnaires allow obtaining data as trips frequency by weekday and schedule, transport mode used, number of
people travelling in the same private vehicle and information relating to the driver destination. They present a
procedure for the implementation of surveys and they suggested follow the Trip Generation Handbook (2001)
guidelines for the schools selection to study. They also propose a specific computer database design in order to
organize the data.

3. Methodology

3.1 Sample selection

The data shown in this study was obtained through surveys carried out in a sample of twenty-one schools, as
specified by the Trip Generation Handbook, according to which, for the trip data, information from at least three to
five study sites should be captured, in order to identify and discard outliers. The selected schools are located in the
Libertador municipality of Mérida, in Venezuela. Private school –non-subsidized– and concerted –subsidized–, with
the aim of establishing a contrast between the trip generations rates are selected. Different data collection methods
may apply, given that outcomes are due to different researchers group. In many schools surveys were applied, and in
some other place, we made vehicle counts by type at access points to the school. The results correspond to data taken
between 2011 and 2013.

3.2 Selection of variables to correlate

Trip Generation Manual presented a major cause of variation variables in the trip generated by schools by
students and employees. It is considered as alternative variables the construction area and the monthly paid by the
fathers. However, both aspects was deleted because generates much uncertainty; in this way, only taking into
account the number of students and employees as independent variables for obtaining generation rates. Student
variable includes the preschool level, basic and middle school, and the employee variable includes teachers,
administrative staff and worker.Here introduce the paper, and put a nomenclature if necessary, in a box with the
same font size as the rest of the paper. The paragraphs continue from here and are only separated by headings,
subheadings, images and formulae. The section headings are arranged by numbers, bold and 10 pt. Here follows
further instructions for authors.

3.3 Data collection

Number of students and employees was obtained through schools head. To obtain the data it is applied surveys in
both variables. The format of the survey includes specific questions.
4 Quintero, Díaz &Moreno/ Procedia - Social and Behavioral Sciences 00 (2014) 000–000

Institution: ____________________________________________________________
Grade and section: ______________________________________________________
Date: _________________________________________________________________
Interviewer: ___________________________________________________________

1. How do you reach the school?

a. private car with a relative, neighbour or friend.
b. school transport.
c. taxi.
d. public transport.
e. walk.
f. driving their own vehicle.
g. motorcycle.

2. When you arrive in private car, taxi, bus or motorcycle, there is more people arriving to same
school in the same vehicle? How many?

3. How do you exit of school?

a. private car with a relative, neighbour or friend.
b. school transport.
c. taxi.
d. public transport.
e. walk.
f. driving their own vehicle.
g. motorcycle.

4. When you come out in private car, taxi, bus or motorcycle, there is more people that use same
vehicle? How many?
Fig. 1 – Survey format

General information is collected, and questions are formulated in order to quantify the trips made by people with
a same origin-destination pattern and to move together in the same vehicle, trying avoiding data redundancy. In the
initial phase of sampling where surveys were applied to private schools, the motorcycle was not a clear mode choice,
however, given the increased use of this mode were applied in concerted schools. Survey format was directly
delivered for the respective filling, while smaller students with difficulty in processing information, were grouped
for the questions formulation orally. Interviewer in this case registered responses from raised hands; in this process
the teacher collaboration was key.

Sampling contents 60 % of the population in each variable, and is selected, through the generation of random
numbers. In some data through vehicle counts are taken in access to school, at entry and exit hours. This technique
allowed quantifying the trips in private vehicles, on school transport, taxi, motorcycle and on foot; trips on public
transport have disadvantage, because stops are out of reach of the counts, therefore, trips this mode could have been
quantified as trips on foot.

3.4 Calculation procedure

Once realized the surveys was totalized the number of trips by transport mode. The results obtained for the
sample were expanded to the population and the data was grouped corresponding to concerted and private schools.
The expansion was made through equation 1.
 Quinterot, Díaz & Moreno/ Procedia - Social and Behavioral Sciences 00 (2014) 000–000 5

Once realized the surveys was totalized the number of trips by transport mode. The results obtained for the
sample were expanded to the population and the data was grouped corresponding to concerted and private schools.
The expansion was made through equation 1.

i  P
Ti  (1)
m

Nomenclature

Ti expected trips by mode i, full students and employees of the school

Ti number of trips by mode i, of the pupils and employees sample
P population, students and employees of the school
m sample size, students and employees of the school

Percentage distributions of the mode are used in order to make comparisons and determine an intensity model by
mode. Each rates of school trip generation were calculated by the ratio between the total number of trips by mode
and the independent variable, which was considered the number of students and employees by school as more
reliable parameters.

Trip generation rate was obtained for each mode on each institution by the equation 2.

Ti
TGRi  (2)
IV

Nomenclature

TGRi trip generation rate by model i for school

IV independent variable associated with the school

Average rates pondered by independent variable with its respective statistical analysis were calculated separately
for concerted and private schools.

TGR  IV
j 1
i

TGRi  Pond  (3)

 IV

Nomenclature

TGRi  Pond average pondered trip generation rate to the mode i

n school number

Total trips in private vehicles, school and taxi transport, divided between the number of students and employees
as indicated above determine the trip generation rates, of this total excludes trips made by public transport because it
6 Quintero, Díaz &Moreno/ Procedia - Social and Behavioral Sciences 00 (2014) 000–000

is considered that these are carried out independently. The results could be expressed by peak hour, in the entry or
exit of school moment. Rates for public transport, on foot and by motorcycle are also obtained, but are considered as
less relevant in the vehicular impact, thus are not detailed here. Total trips in private vehicles, including school bus
and taxi, were plotted for each school versus students, employees, and students more employees. From these graphs,
linear regression equations were obtained.

y  a x (4)

Nomenclature

y number of trips in vehicle/hour

a correlation coefficient
x estimated independent variable

4. Data statistical analysis

A way to validate the obtained average rates is to apply appropriate statistical tests to the data, allowing inference
from the sample to the population to be made. Atypical data is purified with special tests such as Grubbs' test and
Dixon's test, based on approximate intensity to more reliable values with less influence of data that not correspond to
a normal distribution. In this case there are few outliers that are purged to ensure a more consistent analysis.
Satisfactory results allow visualizing a data normal trend, despite the apparent variability, considering all most
influential modes in the street congestion around the schools, are obtained.

Table 1. Overview of the most important statistical parameters in each sample

Sample Average Median Geometric mean Standard Variability
deviation Coefficient
*
M Student CS 9 0.0277778 0.03 0.0258667 0.00971825 34.9857%
M Employees CS 9 0.318889 0.32 0.263095 0.176737 55.4229%
PT Student CS 10 0.337 0.325 0.296594 0.165197 49.0198%
PT Student PS** 10 0.185 0.19 0.152568 0.108551 58.6763%
PT Employees CS 10 2.932 2.91 2.76604 0.983054 33.5285%
PT Employees PS 10 1.751 1.89 1.39417 1.04113 59.4589%
PV Student CS 10 0.601 0.57 0.58436 0.147832 24.5978%
PV Student PS 11 0.826364 0.87 0.804353 0.191221 23.14%
PV Employees CS 10 6.656 5.76 5.53694 3.73124 56.0582%
PV Employees PS 11 8.04727 8.05 7.49649 2.82728 35.1334%
(*) concerted school; (**) private school

The overview shows various statistics for each of the ten data columns. Verification of significant mean
differences must be graphing for a better appreciation of differences between data groups.

Normality tests indicate that obtained rates can be used in the proposed prediction models to determine the
vehicle intensity generated by each type of school -private and concerted-. Case of the motorcycle mode students on
concerted schools is out of the range expected as a normal condition only. However, in the employees sample for
this same school, it is within of the range expected as a normal condition.
 Quinterot, Díaz & Moreno/ Procedia - Social and Behavioral Sciences 00 (2014) 000–000 7

M Employee Concerted Box-and-Whisker Plot

M Student Concerted
PT Employee Concerted
PT Employee Private
PT Student Concerted
PT Student Private
PV Employee Concerted
PV Employee Private
PV Student Concerted
PV Student Private

0 2 4 6 8 10 12
TGR

Fig. 2. Comparison of trip average rates by mode, variable and school

Table 2. Tests of normality in each analyzed class

Chi-square Shapiro Wilk (W)
Test statistic p-value Test statistic p-value
M Student CS 16.0 0.013754 0.90379 0.270184
M Employee CS 4.0 0.676677 0.962341 0.816672
PT Student CS 6.0 0.539749 0.964865 0.828642
PT Student PS 8.0 0.332594 0.9409 0.546575
PT Employee CS 14.0 0.0511814 0.888322 0.154744
PT Employee PS 4.0 0.779777 0.95786 0.74764
PV Student CS 6.0 0.539749 0.95511 0.714609
PV Student PS 9.90909 0.193784 0.917708 0.286152
PV Employee CS 4.0 0.779777 0.93559 0.488619
PV Employee PS 6.27273 0.50829 0.948524 0.602742

This phase of the analysis shows that all cases can be adequately modeled by a normal distribution. The Chi-
square test discriminates against the existing range among the ten probable samples. The Shapiro-Wilk test is based
on the comparison of each sample and its adjustment to a normal distribution. The lowest values of p-value are more
than 0.05 in the majority of cases, what determines that these data come from a normal distribution with a
confidence level of 95 %. Measures of central tendency, variability, and form, of particular interest in this analysis
also determined that the samples come from a normal distribution. As the sharpness and kurtosis sadistic values,
experience ranges between - 2 and 2, indicating significant coming from a normal distribution.

5. Results

In the results you can see that the population percentage surveyed was not exactly the 60 % were treated as the
initial goal. However, the displayed percentage values are calculated by reference to the number of persons actually
surveyed. Surveys applied have totalized the amount of students and employees using every transport mode, these
results were extrapolated to the total population, whereas the trends observed for the sample are applicable to the
population.
8 Quintero, Díaz &Moreno/ Procedia - Social and Behavioral Sciences 00 (2014) 000–000

Table 3 – Distribution of the mode use in each school

Type School Students and employees who use
PV ST Taxi PT Walk M
U.E.P. Nuestra Sra. Rosario 613 150 18 63 65 -
U.E.P. La Presentación 469 158 12 25 20 -
Fundación Monseñor Bosset 268 39 9 27 37 -
La Salle 583 21 6 123 44 -
Teresa Titos 394 6 1 44 18 -
Private

U.E. Dr. Rafael Antonio Uzcátegui 163 18 5 64 31 -

U.E.P. La Presentación 701 4 15 SR 39 4
U.E.P. Arzobispo Silva 296 2 8 138 138 -
U.E.P. Dr. Carlos E. Muñoz Oraá 260 5 5 93 27 -
U.E.C. Prof. Gustavo Garrido 207 15 1 67 74 -
U.E.P. Francisco Lazo Martí 7 0 1 55 21 -
U.E.P. Martin Luther king 80 0 8 67 47 -
Total 4041 418 89 766 561 4
U.E. Micaeliano 118 5 10 169 84 8
U.E. Salesiano San Luis 545 105 15 121 54 28
U.E. Madre Emilia 305 54 13 61 122 20
U.E. Timoteo Aguirre Pe 92 98 1 163 42 1
Concerted

U.E. San Javier del Valle 42 28 22 105 41 4

U.E. Sagrada Familia 185 23 11 159 118 12
U.E. San José de la Sierra 174 22 13 208 50 6
U.E. La Salle Hermano Luis 202 56 30 189 51 20
U.E. Nuestra Sra. de Fátima 424 131 18 123 38 15
U.E. Nuestra Señora de Belén 399 226 19 297 96 26
Totales 2486 748 152 1595 696 140

From the shown values the use distributions of each transport mode is obtained.

80 75
70 Private Concerted
60 48
50
40 32
30 14 15
20 8
10 2 3 3
0
Taxi

Public Transport

Motorcycle
School transport
Particular vehicle

Fig. 3. Motorized transport mode distribution in concerted and private school

Mode distribution shows that the mode most used both in private and concerted school is the private car followed
by public transport. In private school, the percentage of use of the private car is on the order of 70 % while that in
concerted school is around 40 %. Shown the distribution of motorized mode also, which saw as predominantly the
private car with a 75 % for private school and a 48 % for concerted. In particular, there is difference in the use of
 Quinterot, Díaz & Moreno/ Procedia - Social and Behavioral Sciences 00 (2014) 000–000 9

mode in both schools, private and concerted. Trip generation rates are estimated using the total number of trips in
private vehicles, school transport and taxi, altogether.

Table 4 - Particular vehicle trip generation rates

trip generation rates of vehicle particular
School Type
(Trips/student/h) (Trips/employee/h)
U.E.P. Nuestra Sra. Rosario 0,93 11,83
U.E.P. La Presentación 1,03 9,69
Fundación Monseñor Bosset 1,04 4,21
La Salle 0,87 8,13
Teresa Titos 0,94 11,15
U.E. Dr. Rafael Antonio Uzcátegui 0,74 6,17

Private
U.E.P. La Presentación 1,03 11,06
U.E.P. Arzobispo Silva 0,56 8,05
U.E.P. Dr. Carlos E. Muñoz Oraá 0,76 7,70
U.E.C. Prof. Gustavo Garrido 0,67 7,43
U.E.P. Francisco Lazo Martí 0,11 0,73
U.E.P. Martin Luther king 0,51 3,04
Mean pondered rate 0,85 8,18
Standart deviation 0,27 3,42
U.E. Micaeliano 0,36 5,52
U.E. Salesiano San Luis 0,83 10,22
U.E. Madre Emilia 0,68 11,98
U.E. Timoteo Aguirre Pe 0,56 3,29

Concerted
U.E. San Javier del Valle 0,55 1,23
U.E. Sagrada Familia 0,48 4,15
U.E. San José de la Sierra 0,50 4,02
U.E. La Salle Hermano Luis 0,58 6,00
U.E. Nuestra Sra. de Fátima 0,82 11,68
U.E. Nuestra Señora de Belén 0,65 8,47
Mean pondered rate 0,64 6,37
Standart deviation 0,15 3,73

Thus, the estimation models for the trips per hour by school and independent variable are obtained.

Table 5 - Linear model in the estimation of trip generation rate

School Independent variable (x) Coefficient a R2 correlation coefficient
Private Students 0,8901 0,90151
Private Employees 8,4219 0,66688
Private Students + employees 0,8116 0,90602
Concerted Students 0,6728 0,86230
Concerted Employees 6,1530 0,04496
Concerted Students + employees 0,6165 0,84061

For private schools the variable that best describes the number of trips is the number of students plus employees,
while for concerted school this corresponds to the number of students. Concerted school gets a very poor correlation
when using the employee variable. An illustrative way includes the graphs for the variable most widely used –
students and employees– for both private and concerted schools.
10 Quintero, Díaz &Moreno/ Procedia - Social and Behavioral Sciences 00 (2014) 000–000

a 900 b 900
800 y = 0.8116x 800 y = 0.8901x
700 R² = 0.906 700 R² = 0.90151
600 600
500 500
400 400
300 300
200 200
100 100
0 0
0 500 1000 0 200 400 600 800 1000

Fig. 4. (a) Veh trips (hour) vs (employee + students); (b) Veh trips (hour) vs students. Private school

6. Conclusions

• There are differences in the use of the transport mode between concerted and private schools also in the trip
generation rates. The mode used in both types of schools is the private car followed by public transport; however, the
use of the private car is more significant in the private schools.
• Variable that best explains the number of trips for private schools results the number of students most
employees, while concerted school has better fit if you consider only the number of students. You must investigate
more about this result in function to determine new variables influential in its variability.
• Obtained models may be useful if they have the relevant validity. Vehicles counts in access to similar schools
must allow the models calibration and definition of the prediction degree achieved. Only in this way, the estimated
value of the trips will be reliable when planning the vehicles movement and pedestrians around schools.
• Statistical analysis demonstrates in all cases, the trip generation rates obtained are in accordance with a normal
distribution. Except for motorcycle and students in concerted school, this requires further study.
• Future application of alternative techniques for counts in schools access could provide good results in the
calibration process of the models obtained from the implementation of surveys.

References

Herz M, Galarraga J & Pastor G. (2009). Características de generación y distribución modal de viajes en centros educativos universitarios.
Facultad de Ciencias Exactas, Físicas y Naturales. Universidad Nacional de Córdoba. Available on line at:
http://www.redpgv.coppe.ufrj.br/index.php/es/produccion/articulos-cientificos/2009-1/431-caracteristicas-de-generacion-y-distribucion-
modal-de-viajes-en-centros-educativos-universitarios/file. [Cited on 05 January 2013].
ITE – Institute of Transportation Engineers (2001). Trip Generation Handbook - An ITE Recommended Practice. 6ª Ed. Washington D.C
Prudêncio M., Ferreira de Souza S. & others (2006). Procedimento para a coleta e registro dos dados obtidos junto a instituições de ensino no
contexto de pólos geradores de viagens. Universidade de Brasília. Available on line at:
http://www.redpgv.coppe.ufrj.br/index.php/es/produccion/articulos-cientificos/2006-1/291-souza-lima-ferreira-jacques-anpet2006/file. [Cited
on 08 January 2013].
Souza, A. & Prudêncio M. (2010). Características de viagens geradas em instituições de ensino médio em brasília. Universidade de Brasília.
Centro Universitário Izabela Hendrix BH-MG y Universidade de Brasília – DF. Available on line at:
http://www.redpgv.coppe.ufrj.br/index.php/es/produccion/articulos-cientificos/2010-1/580-estudo-da-geracao-de-viagens-em-instituicoes-de-
ensino-medio/file. [Cited on 20 January 2013]